Lightweight:
- GPU collection moves to a dedicated worker thread that owns the gfxinfo
handle for the process lifetime. gfxinfo's active_gpu() runs a full NVML
init/teardown (~20ms, blocking) and we were paying it on the async
runtime for every collect — measured at ~80% of the agent's entire
active CPU on a GPU machine. The handle holds Rc<Nvml> (not Send), so a
thread + mpsc/oneshot channel pair confines it; a zero-total-VRAM reply
is treated as a dead session (driver reload) and re-probed.
- journalctl now runs via tokio::process instead of blocking one of the
two runtime workers for the duration of the subprocess.
- TtlCell (state.rs) replaces the four hand-rolled static TTL caches; a
cached negative result now counts as fresh, so hosts with no matching
temp sensor or GPU stop rescanning every request. Single lock+clone on
the GPU cache hit path (was two).
Correctness:
- Process/child CPU times are now microseconds as documented; they were
milliseconds, rendering 1000x too small next to (correct) thread times.
- Non-Linux per-process CPU%% clamps AFTER dividing by core count; a
4-cores-busy process on an 8-core box reported 12.5% instead of 50%.
- Journal timestamps are real RFC 3339 UTC plus an additive timestamp_us
field (sorting is now numeric); the old strings were Debug-formatted
SystemTime mangled by string replace.
- Partition detection uses /sys/block on Linux: whole-disk filesystems on
names like nvme0n1 or zram1 are no longer misclassified as partitions.
One shared parent_disk_name() replaces two inline copies.
- New sampled_at_ms on the metrics payload (additive) records when the
snapshot was actually collected, so clients can compute exact rates
across the agent's TTL cache.
Security/robustness:
- key.pem is created 0600 (was umask default 0644, world-readable) and
pre-1.51 keys are tightened on startup.
- Per-PID detail/journal caches now evict (60s max age, 64 entries max);
they previously grew without bound under PID-walking clients.
- The two per-PID ws handlers collapse into one generic helper.
- /proc/<pid>/stat parsing unified in one comm-safe module.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* chore: update ratatui from 0.28 to 0.30
* style: cargo fmt
* fix: replace manual zero-guarded divisions with checked_div
* fix: collapse nested if into match guard
* style: cargo fmt
* bump crossterm and optimize various types, remove stale code.
* fix windows build
* only show parent level processes on main tui
This check in offers alpha support for per process metrics, you can view threads, process CPU usage over time, IO, memory, CPU time, parent process, command, uptime and journal entries. This is unfinished but all major functionality is available and I wanted to make it available to feedback and testing.
- socktop connector allows you to communicate with socktop agent directly from you code without needing to implement the agent API directly.
- will also be used for non tui implementation of "socktop collector" in the future.
- moved to rust 2024 to take advantage of some new features that helped with refactor.
- fixed everything that exploded with update.
- added rust docs for lib.
Gate Linux-specific imports and fields to avoid Windows dead-code/unused warnings.
Ensure Linux-only proc CPU tracker is not referenced on non-Linux builds.
Here are concise release notes you can paste into your GitHub release.
Release notes — 2025-08-12
Highlights
Agent back to near-zero CPU when idle (request-driven, no background samplers).
Accurate per-process CPU% via /proc deltas; only top-level processes (threads hidden).
TUI: processes pane gets scrollbar, click-to-sort (CPU% or Mem) with indicator, stable total count.
Network panes made taller; disks slightly reduced.
README revamped: rustup prereqs, crates.io install, update/systemd instructions.
Clippy cleanups across agent and client.
Agent
Reverted precompressed caches and background samplers; WebSocket path is request-driven again.
Ensured on-demand gzip for larger replies; no per-request overhead when small.
Processes: switched to refresh_processes_specifics with ProcessRefreshKind::everything().without_tasks() to exclude threads.
Per-process CPU% now computed from /proc jiffies deltas using a small ProcCpuTracker (fixes “always 0%”/scaling issues).
Optional metrics and light caching:
CPU temp and GPU metrics gated by env (SOCKTOP_AGENT_TEMP=0, SOCKTOP_AGENT_GPU=0).
Tiny TTL caches via once_cell to avoid rescanning sensors every tick.
Dependencies: added once_cell = "1.19".
No API changes to WS endpoints.
Client (TUI)
Processes pane:
Scrollbar (mouse wheel, drag; keyboard arrows/PageUp/PageDown/Home/End).
Click header to sort by CPU% or Mem; dot indicator on active column.
Preserves process_count across fast metrics updates to avoid flicker.
UI/theme:
Shared scrollbar colors moved to ui/theme.rs; both CPU and Processes reuse them.
Cached pane rect to fix input handling; removed unused vars.
Layout: network download/upload get more vertical space; disks shrink slightly.
Clippy fixes: derive Default for ProcSortBy; style/import cleanups.
Docs
README: added rustup install steps (with proper shell reload), install via cargo install socktop and cargo install socktop_agent, and a clear Updating section (systemd service steps included).
Features list updated; roadmap marks independent cadences as done.
Upgrade notes
Agent: cargo install socktop_agent --force, then restart your systemd service; if unit changed, systemctl daemon-reload.
TUI: cargo install socktop --force.
Optional envs to trim overhead: SOCKTOP_AGENT_GPU=0, SOCKTOP_AGENT_TEMP=0.
No config or API breaking changes.
Release highlights
Introduced split client/agent architecture with a ratatui-based TUI and a lightweight WebSocket agent.
Added adaptive (idle-aware) sampler: agent samples fast only when clients are connected; sleeps when idle.
Implemented metrics JSON caching for instant ws replies; cold-start does one-off collection.
Port configuration: --port/-p, positional PORT, or SOCKTOP_PORT env (default 3000).
Optional token auth: SOCKTOP_TOKEN on agent, ws://HOST:PORT/ws?token=VALUE in client.
Logging via tracing with RUST_LOG control.
CI workflow (fmt, clippy, build) for Linux and Windows.
Systemd unit example for always-on agent.
TUI features
CPU: overall sparkline + per-core history with trend arrows and color thresholds.
Memory/Swap gauges with humanized labels.
Disks panel with per-device usage and icons.
Network download/upload sparklines (KB/s) with peak tracking.
Top processes table (PID, name, CPU%, mem, mem%).
Header with hostname and CPU temperature indicator.
Agent changes
sysinfo 0.36.1 targeted refresh: refresh_cpu_all, refresh_memory, refresh_processes_specifics(ProcessesToUpdate::All, ProcessRefreshKind::new().with_cpu().with_memory(), true).
WebSocket handler: client counting with wake notifications, cold-start handling, proper Response returns.
Sampler uses MissedTickBehavior::Skip to avoid catch-up bursts.
Docs
README updates: running instructions, port configuration, optional token auth, platform notes, example JSON.
Added socktop-agent.service systemd unit.
Platform notes
Linux (AMD/Intel) supported; tested on AMD, targeting Intel next.
Raspberry Pi supported (availability of temps varies by model).
Windows builds/run; CPU temperature may be unavailable (shows N/A).
Known/next
Roadmap includes configurable refresh interval, TUI filtering/sorting, TLS/WSS, and export to file.
Add Context...
README.md
jasonwitty
jason